In conventional systems, users have access to a plethora of media content. With so much content available, users often require assistance in selecting content, and content providers often require assistance in selecting what content to provide to users. To aid in selecting content, systems often collect data related to what content a user watches (or does not watch), what devices the user watches that content on, etc. and stores this information. Despite collecting and storing vast amount of data on users, users often access content on multiple devices, and data collected and stored on one device is not used when recommending content on another device.